¹ For ‘noumenalism’, see Problems from Kant, p. 11. Otherwise unattributed page references will be to, this book.

² Nonontological views have been suggested by, among others, Graham Bird, D. P. Dryer, H. J. Paton, Lewis White Beck, Gerold Prauss, Arthur Melnick, Robert Pippin and Henry Allison. The ontological-nonontological distinction is mine, not Van Cleve's.

³ I use the Guyer-Wood translation of the first Critique and the usual ‘A/B’ references to the pages of the first and second editions of that work. The appearing-theory/appearance-theory distinction was developed by H. A. Prichard and applied to Kant by Prichard and by Stephen Barker. See ch. 2 of my Kant's Transcendental Deduction (Kluwer: Dordrecht, 1992)Google Scholar, hereafter KTD, which indicates how to develop both one- and two-world ontological readings of Kant. The application of, the distinction to Kant has been challenged, but I think it is correct.

⁴ Kemp Smith and others have read Kant as an ontological phenomenalism Although Kant's talk of ‘appearances’ supports a phenomenalist interpretation of some sort, some (but not all) intentional-object or related views provide nonphenomenalist, two-world ontological accounts (for example, Wilfrid Sellars's and Richard Aquila's positions, pp. 8, 261; and Option III in KTD, pp. 47-52). Throughout this essay I ignore fine details bearing on Kant's views and those of his interpreters.

⁵ Moore and others stress the first problem here. At pp. 8, 183-4, 49-50, 58-9, 124, and 62-70, Van Cleve notes it, the second problem, and some others that arise if Kant is taken as an ontological phenomenalist. For simplicity, I talk throughout of the object of knowledge as spatio-temporal, ignoring the differences between outer and inner objects. In developing Van Cleve's views, I ignore complications not relevant here. 6 See p. 241 and also , Allison, Kant's Transcendental Idealism (New Haven: Yale University Press, 1983), pp. 30–4Google Scholar . Allison distinguishes ‘phenomenal-ism’ (that is, analytical phenomenalism) from ‘Berkeleian idealism’ (ontological phenomenalism) in a way influenced by Jonathan Bennett. He argues that Kant holds neither position.

⁷ Such views are suggested (although not positively endorsed) in work by, among others, H. A. Prichard, Stephen Barker and myself. They also are taken into account by Sellars, Aquila and others (see pp. 12, 261). I propose below (and argue in KTD, ch. 2, and elsewhere) that Kant accepts both one- and two-world ontological views in various places. Prichard and Barker also support this proposal.

⁸ See the pp. 147-8 arguments against Allison's views in Kant's Transcendental Idealism. Here Van Cleve in part follows David Lewis's denial (in a non-Kantian setting) that spatial properties are relations. He also takes from Lewis the options (a) to (c) for using modifiers to remove the present sort of contradiction.

⁹ See pp. 149-50. Besides such merging of distinct appearances into one object in itself, there is also the possibility of one appearance's branching into (being grounded on) distinct objects in themselves. See also KTD, ch. 2 and p. 351, n. 42.

¹⁰ See, for example, , Allison, Kant's Transcendental Idealism, pp. 8, 25, 30, 240-2Google Scholar .

¹¹ See pp. 7-8 and 146-50(and note pp. 150-5), from which I take the Van Cleve points below. Karl Ameriks, Paul Guyer, Kenneth Westphal, I myself (KTD, pp. 19-23, 342-5, 348-9), and others have questioned Allison's view.

¹² For Allison's endorsement, see the passages quoted in KTD, ch. 1, note 38, pp. 344-5. Van Cleve, p. 8, puts the problem nicely: ‘I have shoes on my feet. But consider me apart from my shoes: so considered, am I barefoot? I am inclined to say no; consider me how you will, I am not now barefoot.’ , Guyer (Kant and the Claims of Knowledge, Cambridge: Cambridge University Press, 1987, pp. 337–8)Google Scholar offers a similar criticism of Allison.

¹³ Van Cleve also observes that for Allison to show that it is analytic that things in themselves are not in space and time, he will have to show that ‘it is analytic that space and time are forms of human sensibility’ (and so are a priori conditions on our knowing objects) (p. 259, n. 6). As Van Cleve notes, Kant does not seem to regard that point as analytic.

¹⁴ See the first Critique passages assembled at pp. 6-7 and, for example, A 42/B 59; also KTD, pp. 19-23 and p. 344, n. 32, 33 and 35. Van Cleve in fact makes the point, noted above, about the apparent contradiction in a thing's being both spatiotemporal and nonspatiotemporal, as well as the points about the difficulties for one-world views posed by the Amphibolies, in order to undermine Allison's nonontological one-world view. But they apply equally to what I call such ontological views. As he implies (p. 148), some of his criticisms of one-world relational analyses of shape might be disputed. However, his p. 149 questions about mereology and the reducibility of relations ought to make one-worlders blanch. His pp. 149-50 identity-of-indiscernibles discussion (like the related merging-branching issues indicated above in n. 9) also makes views like Allison's difficult to defend, given Kant's own claims. In addition, Allison's views conflict with the fact that, in various places, Kant stresses the possibility of objects that exist in themselves but do not appear to us in our knowledge (for example, God, on Kant's conception).

¹⁵ See pp. 5 and 34-7. I doubt some of Van Cleve's pp. 34-7 claims. But I agree that Allison's view misconstrues Kant's own rationale for the Copernican Revolution.

¹⁶ Van Cleve's account thus is not a pure reconstruction; he bases it on a skilful appeal to some significant texts. He notes that Kant never sharply separates analytical from ontological phenomenalism (pp. 71,123-4). He also counsels that, given the problems involved in supposing that nonspatiotemporal things in themselves affect the mind, we should retain the idea that things in themselves exist and affect the mind; but we may want to abandon the idea that they are nonspatiotemporal (pp. 138, 164). Elsewhere he suggests a defence of hard-core realism about the objects of knowledge (pp. 142-3). Of course he does not hold that Kant himself accepts such a position.

¹⁷ For Van Cleve's position (and his explicit claim that ‘appearances, as things merely having intentional being, do not really exist,’ p. 50), see especially pp. 11-12, 50-1, 58-9, 91-4, 150 and also 123-4; 281, n. 60; 282, n. 82; 260-1, n. 24. Truths about spatiotemporal objects need not be translatable into truths about cognitive states; they must simply be derivable from such truths, where ‘derivable from’ covers any necessary implication (p. 11).

¹⁸ For Van Cleve's Kant, the cup thus exists no more than does a shadow creeping across a lawn. ‘We may truly say that there is a shadow,’ but so speaking ‘does not commit us ontologically to an aetherial two-dimensional entity that is literally on the move. The whole truth in what we say is exhausted by … facts about the sun, the lawn, and the intervening shade tree (p. 9). ‘[V]irtual objects exist only in a manner of speaking; to say that they exist is just shorthand for saying certain things about the more basic entities out of which they are constructions’ (p. 11). Van Cleve also describes Kantian appearances as intentional objects, taking such objects to be virtual objects (pp. 12,150, 164).

¹⁹ He attributes such a view, though not its application to Kant, to Ernest Sosa (p. 11). He does not explain how this sort of view escapes his earlier Humean complaint (p. 10) that an object distinct from a perceptual state should not be dependent on it. Van Cleve takes the supervenient-entity interpretation sometimes to work better than the virtual-object interpretation (see, for example, p. 260, n. 24).

²⁰ See p. 145. At p. 146 he suggests that Kant's language points both ways, a suggestion with which I agree (see below). It is interesting to note that both Allison and Van Cleve acknowledge Kant's Zweideutigkeit about two- and one-world views; but Allison then plays down the two-world texts and Van Cleve plays down the one-world ones. I think that they both oversimplify Kant's actual, ambiguous position.

²¹ See KTD, ch. 2.1 return to this point below.

²² The fact that, on the virtual-object interpretation, a sentence like ‘the cup exists’ says something true does not, of course, imply that there really exists an entity that is referred to by ‘the cup.’ Van Cleve agrees. (See the texts cited in n. 18 and 19 above and esp. pp. 260-1, n. 24.) He recognizes that Kant stresses the empirical reality of appearances (p. 50). But I do not think Kant would accept his account of such reality as implying that entities like cups are logical constructions and ‘not genuinely existent entities’ (p. 50).

²³ See also B 69-70 and Prolegomena, §13, Notes I to III, and the Appendix.

²⁴ See p. 281, n. 60. I here extend the answer a bit. The present issues are complex and deep. I myself think that the intentionality of intuitions is best seen as deriving from the knower's thought that there is an object that has the features displayed by the manifold that the knower is synthesizing. (See A 247/B 304 – ‘thinking is the action of relating given intuitions to an object’ – and KTD, ch. 8 and also chs 3 and 10.) In cases of genuine empirical knowledge, Kant would take this thought to be true and its contained existential quantifier to be read objectually, which again causes problems for the virtual-object interpretation. It would help to have more discussion by Van Cleve of texts like A 247/B 304, which he mentions only at p. 238. (At pp. 95-7, he does, however, discuss Rolf George's valuable work on the intentionality of intuition.)

²⁵ Note also familiar texts like A 165, B 162; A 182, B 224; and A 189, B 232.

²⁶ Thus Van Cleve substitutes ‘being an appearance entails the occurrence of certain perceptions’ for the more straightforward Kantian claim that ‘an appearance is an entity that would not exist unless it were perceived’ (pp. 260-1, n. 24). One can suggest other logical devices here (for example, substitutional quantification), but none seems to deal adequately with all the sorts of statements noted here and below.

²⁷ The problems about the need to quantify over the objects of knowledge could be avoided by the supervenient-object position. However, that position does not escape the Moore problem. Moreover, if we are forced to that position, then why the detour through the virtual-object view in the first place?

²⁸ The same sort of visual experience of a green apple could result from seeing a green apple under normal conditions or from seeing a yellow apple while wearing sunglasses or from some visual defect; and so on. See Chisholm, Roderick, Theory of Knowledge, 2nd edn (Englewood Cliffs, NJ: Prentice Hall, 1977), p. 127Google Scholar .

²⁹ Such phenomenalism takes material objects to be theoretical entities posited to explain our perceptions.

³⁰ Suppose p is the actual (noumenal) perceptual state that is supposed to determine (or on which is supposed to supervene) the truth of the virtual-object claim ‘the apple is green.’ Then it is not clear why p could not also occur when, as we would say in material-object language, the fact is that the apple is yellow, but it looks green to any perceivers because - in a way currently not perceptible by those perceivers - they suffer from some physical visual disorder. But then this yellow-but-visual-disorder fact (or virtual-object fact, for Van Cleve), which actually does not exist, cannot be ruled out by the occurrence of p itself. So, on the virtual-object approach, some other (noumenal) perceptual state p' must occur that determines the yellow-but-visual-disorder fact actually not to exist. However, p'then seems to yield the same sort of problem. (Although p' supposedly determines that the yellow-etc. fact does not exist, the perceptual state p' could still occur when the yellow-etc. fact does exist and some further anomaly occurs in the conditions of observation or in the perceiver.) So the problem is not resolved. Moreover, and in any case, i t seems there need be no current, actual, perceptual state that rules out the fact that a perceiver suffers a visual disorder that she cannot perceive, and so on. Yet Van Cleve will not be happy about resting the actual nonoccurrence of this fact on merely possible perceptual states (cf. p. 241). But if it does not rest on such possible perceptual states, on what does it rest? On subjunctive facts about what would be perceived were conditions to change in certain ways? But such facts do not look, ontologically, much solider than possible perceptions; and, in any case, past phenomenalist appeals to subjunctive conditionals have run into serious problems.

³¹ See KTD, ch. 2, and earlier work cited there.

³² As noted, Van Cleve, whose sympathies lie with realism, himself suggests in several places that one might want to abandon Kant's doctrine that objects in themselves are nonspatiotemporal. (See n. 17 above and pp. 138, 164.). He does not suggest that he himself favours phenomen-alism as against seeing it as present in Kant's work. I myself favour a reconstruction that appeals to intentionality in a way suggested, nonreconstructively, in KTD and in earlier essays.

³³ See esp. pp. 130-2.

³⁴ For an argument that Kant does not accept the Kant-Frege view, see Hintikka, Jaakko, ‘Kant on existence, predication, and the ontological argument,’ Dialectica, 35 (1981), 127–46CrossRefGoogle Scholar . Van Cleve himself is uneasy about the claim that all beings are necessary beings when it is coupled with others of his results (see pp. 197-8, where he leaves the ultimate resolution up to the reader). He says also that his version of the First Analogy is the best that he can do for that argument (p. 121). It seems wrong to describe him as categorically endorsing everything that he tries out using the Kant-Frege idea.

³⁵ See p. 197. This step infers ‘It is necessary that [(x)Ex → Ea]’ from ‘It is necessary that (x)Ex', on the basis of universal instantiation from ’(x)Ex' to ‘Ea’, where V is a name for a being and ‘Ea’ symbolizes 'a exists’. However, considered semantically, this step (and its use of universal instantiation) cannot be valid unless, at each possible world w, the claim (C) '(x)Ex → Ea' holds true. But ‘(x)Fx’ holds true at a world w just in case each object in w satisfies ‘Fx’. So (C) will be false at any world w in which (i) every object in w exists (that is, satisfies ‘Ex)’ and (ii) at w, the name V names an entity that does not itself exist in w. To assume that there cannot be such names is perilously close to assuming the desired result. Moreover, a name like ‘Socrates’ can indeed be used to say or stipulate, of some given world w, that that name's actual-world bearer does not exist in w. ‘Socrates’ also can be used, in 2003, to say that that bearer does not exist in 2003. Allowing such names may conceivably ‘depart… from the spirit of the Kant-Frege view,’ p. 197. But the fact that such names exist and can be used in this way surely undermines that spirit in any case. So too do the counterintuitive results - that all beings are necessary beings and that absolute existence change is impossible - that Van Cleve argues to follow from it.

³⁶ Considering the detailed relation of the categories to the logical functions also brings out Kant's deep reliance on Aristotelian ideas, a reliance that, as argued in detail in KTD, chs 10 and 3, underlies (and explains) crucial steps in the Deduction as Kant presents it. For Van Cleve on the categories and logical functions, see pp. 88-9.

³⁷ I might note another idiosyncrasy here. Probably because of the demands of finishing his work, Van Cleve has stopped textual citations at around 1992, with a few stray references to later publications. As a result, he does not note my 1992 book on the Deduction, which makes a number of the points he sees as crucial to the Deduction's assessment. Nor do works like Brook's, AndrewKant and the Mind (Cambridge: Cambridge University Press, 1994)CrossRefGoogle Scholar and Falkenstein's, LomeKant's Intuitionism (Toronto: University of Toronto Press, 1995)Google Scholar, which also are relevant to his project, get remarked.

³⁸ SeeCh. 1, pp. 12-14.

³⁹ See pp. 217-25 and , Posy, ‘Dancing to the antinomy: a proposal for transcendental idealism,’ American Philosophical Quarterly 20 (1983), 81–94Google Scholar ; ‘Kant's mathematical realism,’ The Monist 67 (1984), 115–34CrossRefGoogle Scholar .

⁴⁰ Arthur Melnick also argues that Kant does not deny (the relevant applications of) excluded middle. See his Space, Time, and Thought in Kant (Kluwer: Dordrecht, 1989), pp. 383–4.Google Scholar